The standard cost comparison (levelized cost of energy) charges every power source for its construction, fuel, operations, and financing — but charges nobody for what goes into the air. This page adds that missing column. Drag the carbon price and watch which technologies were only ever cheap because the atmosphere wasn't sending invoices.
Three things to notice as you drag. First, wind and solar barely move — their red segments at $190/ton are $2 and $9, because lifecycle analysis charges them only for manufacturing. Second, the order reshuffles: coal falls behind nuclear at about $41/ton once health costs are on; gas combined cycle — the "cheap" option — pulls even with nuclear near the EPA's own central carbon estimate. Third, no slider position exists at which fossil fuels win. The debate is only about how much they lose by.
Every number on this page is stated below with its source and arithmetic, so you can disagree with an input and re-derive the result. Where judgment was required, the choice flatters fossil fuels: LCOE averages rather than range extremes, lifecycle emissions charged to renewables too, and health costs set at the conservative end of the published range.
The blue segments are the 2024 average unsubsidized levelized cost of energy: total lifetime cost (construction, fuel, operations, maintenance, financing, and decommissioning for nuclear) divided by total lifetime output. Values are the 2024 endpoints of the University of Michigan Center for Sustainable Systems chart, which averages the high and low of each range in Lazard's annual LCOE analysis. No subsidies or tax credits are included for any source.
| Source | LCOE ($/MWh) | Lifecycle CO₂ (t/MWh) |
|---|---|---|
| Wind (onshore) | $50 | 0.011 |
| Solar (utility PV) | $60 | 0.048 |
| Gas (combined cycle) | $76 | 0.49 |
| Geothermal | $85 | 0.038 |
| Coal | $118 | 0.82 |
| Gas (peaking) | $169 | 0.65 |
| Nuclear | $182 | 0.012 |
The red segment is simply lifecycle emissions multiplied by the selected carbon price:
Emissions intensities are lifecycle medians in line with the IPCC's assessment literature — they include fuel combustion AND upstream effects: mining, manufacturing, transport, and methane leakage for gas. This is why solar and wind carry nonzero values (0.048 and 0.011 t/MWh): their manufacturing footprint is charged, not waived. Nobody rides free on this chart.
The carbon price presets are the two most-cited US federal figures for the social cost of carbon — the estimated economic damage caused by emitting one additional ton of CO₂ (crop losses, property damage, health impacts, and other quantified climate damages). $51/ton was the federal interim central value used through the early 2020s; $190/ton is the EPA's 2023 final central estimate (2% near-term discount rate). Reasonable economists argue for values below, between, and well above both — hence the slider. Pick your own.
Burning fuel emits more than CO₂: fine particulates, sulfur dioxide, and nitrogen oxides cause asthma, heart disease, and premature death in downwind populations. These damages have been monetized for decades (National Academies, Hidden Costs of Energy, 2010; the EU's ExternE project). When the checkbox is on, this page adds conservative per-MWh values for combustion sources only:
The National Academies' 2010 study put average US coal damages near $32/MWh in 2007 dollars (about $45+ today), with a wide plant-by-plant range. Emissions controls and retirements have lowered the fleet average since; $30 splits the difference conservatively. Gas values reflect its far lower particulate and SO₂ output. Non-combustion sources produce no stack emissions, so they receive no adder — their manufacturing pollution is already inside the lifecycle carbon figure.
Fair warning up front: this is the least certain number on the page, and it is included because the question is legitimate, not because science can answer it precisely. Published ecosystem valuations span an order of magnitude or more. The carbon and health segments are literature-grade; this segment is a transparent model with mid-range inputs. Treat it as "roughly this size," never "exactly this number."
The model prices each source's habitat disturbance the same way for everyone:
| Source | Adder ($/MWh) | Plausible range | What it prices |
|---|---|---|---|
| Coal | $15 | $2–60 | Mine-site habitat loss with decades-long persistence, mountaintop-removal ecosystem losses, buried streams, acid drainage habitat damage (Epstein et al. 2011 full-cost accounting) |
| Gas (CC & peaking) | $5 | $1–15 | Well-pad and pipeline habitat fragmentation across producing basins; fragmentation effects extend beyond the cleared footprint |
| Solar (utility PV) | $3 | $0.5–8 | Land conversion at NY capacity factor, partial quality loss (ground cover and pollinator habitat often persist under panels); drops sharply for brownfield and agrivoltaic siting |
| Wind (onshore) | $1 | $0.2–4 | Direct pad/road footprint (~2–5% of spanned area) plus monetized bird and bat mortality per MWh |
| Geothermal | $1 | $0.5–3 | Plant and well-field footprint |
| Nuclear | $0.5 | $0.1–2 | Smallest land footprint per MWh of any source; uranium mine habitat included |
Why the numbers are small. This surprises people on both sides: when habitat damage is spread across the enormous energy output of a power plant, even generous ecosystem valuations yield single-digit dollars per MWh. The European ExternE project and the National Academies reached the same conclusion — climate and human-health damages dominate the externality bill; ecosystem line items are real but comparatively small. The green segment being thin is not a thumb on the scale. It is the finding.
Two honest footnotes. First, wind's wildlife toll is the most publicized item in this category, so for scale: US studies put wind-turbine bird mortality orders of magnitude below deaths from buildings, vehicles, and cats — and fossil generation kills more birds per GWh than wind once mining, pollution, and climate effects are counted. Second, the largest projected driver of biodiversity loss this century is climate change itself, which means much of the "true" biodiversity cost of fossil fuels lives inside the red carbon segment, not this green one. Double-counting is avoided here by keeping this segment to direct, physical habitat effects only.
| Excluded | Would mostly raise the cost of… |
|---|---|
| Firming / storage costs for intermittency | Wind and solar (roughly $40–50/MWh in some regions per Lazard's firming analysis — the one legitimate adder that cuts against renewables; even applied in full, wind and solar remain below gas at the EPA carbon price) |
| Water contamination, acid mine drainage, coal-ash ponds | Coal, gas |
| Methane leakage above the assumed lifecycle rate | Gas (recent satellite surveys suggest official leakage figures are undercounts) |
| Biodiversity and ecosystem damages beyond direct habitat effects | Direct habitat disturbance is now available as the optional green segment (see its panel and its warnings). Broader ecosystem cascades resist honest monetization — and the largest single driver, climate change itself, is already priced inside the carbon segment. |
| End-of-life recycling credits | Would lower solar and wind (glass, aluminum, steel recovery) |
| Nuclear waste storage beyond decommissioning | Nuclear (small per-MWh; US plants already pay into a federal waste fund) |
The one exclusion that genuinely favors renewables on this chart is firming. It is listed first, in bold daylight, because a comparison you can defend is worth ten you can't.